#1870: memory leak on sending management frames
------------------------------------+---------------------------------------
      Reporter:  [EMAIL PROTECTED]  |       Owner:  mentor
          Type:  defect             |      Status:  new   
      Priority:  major              |   Milestone:        
     Component:  madwifi: driver    |     Version:  trunk 
    Resolution:                     |    Keywords:        
Patch_attached:  0                  |  
------------------------------------+---------------------------------------
Old description:

> I was testing source:madwifi/[EMAIL PROTECTED] on an IXP425 in an Avila from
> Gateworks, www.gateworks.com/products/avila/gw2348-4.php.  I am using
> kernel version 2.6.18.4 on ARM in big endian mode.  I have two mini-PCI
> radios, an XR2 and an XR5 from Ubiquiti. See
> wiki:Compatibility/Ubiquiti#XtremeRange2 and
> wiki:Compatibility/Ubiquiti#XtremeRange5.  Both are based on the 5212.
>
> I have two VAPs created as APs.  I noticed over time my free memory kept
> decreasing.  No applications had high usage in top.  A bisectional search
> lead me to [3401].  This changeset sets the bf_node and bf_skb and takes
> a reference on the bf_node.  The problem is that it then calls
> ath_tx_start which does the same starting at line 7543. See
> source:madwifi/trunk/ath/[EMAIL PROTECTED] .
>
> This leads to an extra reference being taken that is never removed.

New description:

 I was testing source:madwifi/[EMAIL PROTECTED] on an IXP425 in an
 [http://www.gateworks.com/products/avila/gw2348-4.php Avila from
 Gateworks].  I am using kernel version 2.6.18.4 on ARM in big endian mode.
 I have two mini-PCI radios, an [wiki:Compatibility/Ubiquiti#XtremeRange2
 XR2] and an [wiki:Compatibility/Ubiquiti#XtremeRange5 XR5] from Ubiquiti.
 Both are based on the 5212.

 I have two VAPs created as APs.  I noticed over time my free memory kept
 decreasing.  No applications had high usage in top.  A bisectional search
 lead me to r3401.  This changeset sets the bf_node and bf_skb and takes a
 reference on the bf_node.  The problem is that it then calls ath_tx_start
 which does the same starting at
 [source:madwifi/trunk/ath/[EMAIL PROTECTED] line 7543].

 This leads to an extra reference being taken that is never removed.

-- 
Ticket URL: <https://madwifi.org/ticket/1870#comment:3>
madwifi.org <http://madwifi.org/>
Multiband Atheros Driver for Wireless Fidelity
-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Madwifi-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/madwifi-tickets

Reply via email to